September 12, 1912] 



NATURE 



primeval structure and have underg-one only slight 

 modification, the other, to which the author 

 belongs, believing that land and sea have changed 

 places at various geological periods. The latter 

 vievi', he considers, is uplield by a consideration 

 of the distribution of geological formations on both 

 sides of the North Atlantic. 



.\ detailed description is then given of the sub- 

 merged river-valleys occurring off western Europe 

 and Africa and in the Mediterranean as traced from 

 the soundings shown on the charts. These were 

 formed not only by rivers, the greater part of 

 the course of which is visible on land, such as the 

 Loire and the Congo, but also by rivers which rose 

 on land now completely submerged, such as the 

 " Irish Channel River " and the " English Channel 

 River." They all indicate a former great uplift 

 of land. The Norwegian fjords also are regarded 

 as river-valleys of great geological age. 



Professor Spencer shows that the continental 

 shelf off the east coast of America is likewise cut 

 up by submarine river-valleys and that there was 

 a land connection between the West Indies and 

 the American continent, and he upholds the view 

 that great changes of level, amounting in some 

 cases to thousands of feet, have taken place in 

 recent geological times. 



In the final chapter Prof. Hull gives his explana- 

 tion of the cause of the Glacial Period. As shown 

 from a study of the submerged river-valleys, a 

 general elevation of the earth's crust took place 

 all round the North Atlantic, the date of which 

 is concluded to be about the close of the Tertiary- 

 Period. This brought about a much colder 

 climate and at the same time a great change in 

 the direction and temperature of the Gulf Stream. 

 When the Antilles were directly connected with 

 the American continent this current could not 

 enter the Caribbean Sea, where at present it gains 

 about 13° Fahrenheit of temperature; hence arose 

 an additional cause for decreased temperature along 

 all the coasts of the North Atlantic. The com- 

 bined effect of these two factors, viz. , the increased 

 elevation of land on both sides of the Atlantic and 

 the decrease of temperature in the Gulf Stream, 

 would be sufficient, the author considers, to call 

 into existence a rigorous glacial climate over the 

 northern parts of America and Europe, which in 

 its turn would affect a great part of the rest of 

 Europe and western Asia, and more or less the 

 entire northern hemisphere. Thus Dr. Hull shows 

 that he belongs to those who regard purely terres- 

 trial factors as the cause of the Glacial Period, 

 in contrast to those who explain it on an astro- 

 nomical basis. The book is useful to all who 

 are interested in physical geography, whether they 

 can agree with Dr. Hull's conclusions or not. 

 NO. 2237, VOL. 90] 



OUR BOOKSHELF. 



The Elements of Statistical Method. By 



Willford I. King. Pp. xvii-f-250. (New York': 

 The Macmillan Company ; London : Macmillan 

 and Co., Ltd., 1912.) Price bs. 6d. net. 

 In this volume Mr. King has endeavoured " to 

 furnish a simple text in statistical method for the 

 benefit of those students, economists, administra- 

 tive officials, writers, or other members of the 

 educated public who desire a general knowledge 

 of the more elementary processes involved in the 

 scientific study, analysis, and use of large masses 

 of statistical data." After a brief historical in- 

 troduction, he outlines the uses and sources of 

 statistical data, and then gives a few short 

 chapters on "the gathering of material"; the 

 third part, forming the bulk of the book, deals 

 with " analysis," i.e., tabulation, averages, dis- 

 persion, correlation, and so forth. 



The writing of a satisfactory elementary work 

 on such a subject — a work that can be placed in 

 the hands of the junior student with confidence 

 that he will not have to unlearn at a later stage 

 some of the notions that he has gathered — is an 

 exceedingly difficult feat, much more difficult in 

 many respects than the ^writing of a work for 

 more advanced students, and we cannot say that, 

 in our opinion, Mr. King has altogether succeeded. 

 The style is simple enough, but some matters are 

 very insufficientlj' explained — probable errors, for 

 example — and in other cases, notably in the 

 chapters dealing with correlation, extensive re- 

 vision and correction are required. A coefficient 

 "of concurrent deviations" suggested on p. 208 

 does not fulfil the fundamental condition of be- 

 coming equal to zero if the deviations are inde- 

 pendent. The student, in dealing with correlation, 

 is repeatedly told to divide deviations by the mean, 

 and the graph of regression obtained when the 

 deviations have been divided by their respective 

 means is termed the " Galton graph." What Sir 

 Francis Galton did was to divide deviations by 

 their respective quartile deviations — not their 

 means — and he obtained the correlation coefficient 

 graphically in that way. The relation of regres- 

 sion to correlation is never clearly exhibited, and 

 Mr. King's use of the term is not in accordance 

 with general usage. As it at present stands, the 

 book cannot be recommended as a completely trust- 

 worthy guide. 



.inthropologie Anatoiniqiie. Crane — Face — Tete 

 sur le Vivant. By Dr. Georges Paul-Boncour. 

 Pp. xix + 396. (Paris: Octave Doin et Fils, 

 1912.) Price 5 francs. (Encyclopedic Scien- 

 tifique.) 

 The enterprising publishers of the " Encyclopedic 

 Scientifique " have arranged for the issue of a 

 series of forty-eight volumes dealing with anthro- 

 pology, the editorship of the series being assigned 

 to Prof. Papillault, of 1' cole d 'Anthropologic, 

 Paris. This volume, by Dr. Georges Paul- 

 Boncour, forms the first of the series, and if its 

 successors maintain an equally high standard, the 

 " Biblioth^que d 'Anthropologic " is destined to 

 become a standard work. 



